Aiming at the problems of insufficient real-time performance and low positioning accuracy in the leakage diagnosis of heating pipe network in Inner Mongolia Shuangxin Mining Co., Ltd., this paper proposes a comprehensive grading heating pipe network leakage diagnosis method combining CUSUM algorithm and BP neural network algorithm. This method constructs a real-time leakage diagnosis and positioning system by combining CUSUM algorithm and BP neural network. Firstly, based on the real-time monitoring data of the secondary network 's make-up water flow, the CUSUM algorithm is combined with the simulation model of the heating pipe network to realize the first-level diagnosis of leakage occurrence and leakage. Subsequently, combined with the pipeline network operation data and simulation model data, the BP neural network algorithm is used to perform secondary diagnosis of the leakage location. The application effect of the system shows that the leakage / non-leakage accuracy and leakage location detection accuracy of the No.3 building heat exchange station, the auxiliary wellhead heat exchange station and the boiler room heat exchange station of Shuangxin Mining Co., Ltd.all reach 100 %. The system response delay time is less than 2 minutes, and the average response time is less than 1 minute. The research results provide a new solution for the intelligent leakage diagnosis of industrial heating pipe network, which has important practical application value and has positive reference significance for ensuring heating safety and reducing energy consumption.